1 Introduction

This paper contains estimates for the effective reproduction number \(R_{t,m}\) over time \(t\) in various provinces \(m\) of South Africa. This is done using the methodology as described in [1]. These have been implemented in R using EpiEstim package [2] which is what is used here. The methodology and assumptions are described in more detail here.

This paper and it’s results should be updated roughly daily and is available online.

As this paper is updated over time this section will summarise significant changes. The code producing this paper is tracked using Git. The Git commit hash for this project at the time of generating this paper was bd94775cd5707ed7cb4139149037a5f63b196b12.

A major recent update made on 29 March was to switch from data based on cases reported as captured in [3] to a data source that contains cases by date the specimen was received [4].

2 Data

Case data is extracted from the NICD National COVID-19 Daily Report [4]. This contains the daily cases reported by the NICD for South Africa by province. Data is shown by specimen reported date. Most recent data is excluded due to incomplete reporting of tests in last number of days.

This report contains data as released at 2021-04-05 21:00:59.

The following fixes are applied:

  1. Calculate daily new cases from cumulative data captured.
  2. Add a total for South Africa as a whole.
  3. Add records (with 0 case count) in periods where no cases were recorded.
  4. Recalculate cumulative figures.
  5. Data captured for specimens received in the 3 days before the reporting date is excluded due as cases reporting is likely incomplete.

3 Methodology

The methodology is described in detail here.

4 Results

4.1 Cases

Below a 7-day moving average daily case count is plotted by province on a log scale since start of the epidemic:

Daily Cases by Province (7-day moving average)

Daily Cases by Province (7-day moving average)

Below the above chart is repeated for the last 30-days:

Daily Cases for Last 30-days by Province (7-day moving average)

Daily Cases for Last 30-days by Province (7-day moving average)

4.2 Current \(R_{t,m}\) estimates by Province

Below current (last weekly) \(R_{t,m}\) estimates are tabulated.

Estimated Effective Reproduction Number by Province
province Count (Week) Week Ending Reproduction Number [95% Confidence Interval]
Eastern Cape 152 2021-04-01 1.2 [1.01 - 1.41]
Free State 843 2021-04-01 1.0 [0.97 - 1.12]
Gauteng 2,021 2021-04-01 1.0 [0.97 - 1.07]
KwaZulu-Natal 804 2021-04-01 1.1 [1.00 - 1.18]
Limpopo 171 2021-04-01 0.9 [0.74 - 1.00]
Mpumalanga 945 2021-04-01 1.1 [1.02 - 1.18]
North West 707 2021-04-01 1.1 [0.98 - 1.14]
Northern Cape 519 2021-04-01 1.0 [0.90 - 1.08]
Western Cape 1,034 2021-04-01 1.0 [0.96 - 1.10]
South Africa 7,196 2021-04-01 1.0 [1.00 - 1.08]
Estimated Effective Reproduction Number by Province

Estimated Effective Reproduction Number by Province

4.3 Maps of Effective Reproduction Number

Below estimates of the reproductive number are plotted on a map of South Africa [5].

Estimated Effective Reproduction Number Based on Cases by Province

Estimated Effective Reproduction Number Based on Cases by Province

4.4 Estimated Effective Reproduction Number for South Africa over Time

Below the results for South Africa over the last 90 days are plotted together with a plot since start of the pandemic.

Estimated Effective Reproduction Number Based on Cases for South Africa over last 90 days

Estimated Effective Reproduction Number Based on Cases for South Africa over last 90 days

Estimated Effective Reproduction Number Based on Cases for South Africa since 1 April 2020

Estimated Effective Reproduction Number Based on Cases for South Africa since 1 April 2020

4.5 Map of Effective Reproduction Number Over Last 60 Days

Below the reproduction number by week by province is animated:

4.6 Estimated Effective Reproduction Number for Provinces over Time

The results for each province over last 90 days are plotted below.

4.6.1 Eastern Cape

Estimated Effective Reproduction Number Based on Cases for Eastern Cape over Time

Estimated Effective Reproduction Number Based on Cases for Eastern Cape over Time

4.6.2 Free State

Estimated Effective Reproduction Number Based on Cases for Free State over Time

Estimated Effective Reproduction Number Based on Cases for Free State over Time

4.6.3 Gauteng

Estimated Effective Reproduction Number Based on Cases for Gauteng over Time

Estimated Effective Reproduction Number Based on Cases for Gauteng over Time

4.6.4 KwaZulu-Natal

Estimated Effective Reproduction Number Based on Cases for KwaZulu-Natal over Time

Estimated Effective Reproduction Number Based on Cases for KwaZulu-Natal over Time

4.6.5 Limpopo

Estimated Effective Reproduction Number Based on Cases for Limpopo over Time

Estimated Effective Reproduction Number Based on Cases for Limpopo over Time

4.6.6 Mpumalanga

Estimated Effective Reproduction Number Based on Cases for Mpumalanga over Time

Estimated Effective Reproduction Number Based on Cases for Mpumalanga over Time

4.6.7 Northern Cape

Estimated Effective Reproduction Number Based on Cases for Northern Cape over Time

Estimated Effective Reproduction Number Based on Cases for Northern Cape over Time

4.6.8 North West

Estimated Effective Reproduction Number Based on Cases for Gauteng over Time

Estimated Effective Reproduction Number Based on Cases for Gauteng over Time

4.6.9 Western Cape

Estimated Effective Reproduction Number Based on Cases for Western Cape over Time

Estimated Effective Reproduction Number Based on Cases for Western Cape over Time

4.7 Detailed Results

Detailed output for all provinces are saved to a comma-separated value file. The file can be found here.

5 Discussion

Limitation of this method to estimate \(R_{t,m}\) are noted in [1]:

  • It’s sensitive to changes in transmissibility, changes in contact patterns, depletion of the susceptible population and control measures.
  • It relies on an assumed generation interval assumptions.
  • The size of the time window can affect the volatility of results.
  • Results are time lagged with regards to true infection.
  • It’s sensitive to changes in case detection.
  • The generation interval may change over time.

For example the the generation interval is based on international data and not South African data.

Further to the above the estimates are made under assumption that the cases are detected at a similar ratio to the underlying infections over time. Should this change rapidly over an interval of a few weeks the above estimates of the effective reproduction numbers would be biased. For example a rapid expansion of case detection could result in an overestimating recent effective reproduction numbers. A more practical example may be a public holiday resulting in fewer specimens received on a particular day. Or, also, changes in testing, such as using more antigen tests (with lower sensitivity) relative to PCR tests could reduce the case ascertainment rate per test for example.

Estimates for the reproduction number here are plotted in time period in which the specimen was received. Ideally we would wish to capture the date the infection occurred. These figures have not been shifted back, even approximately, to the date of infection.

Despite these limitation it is believed that the ease of calculation of this method and the ability to use multiple sources makes it useful as a monitoring tool.

6 Comparison with other estimates

In [6] similar estimate of the reproduction number are made. The main differences from those estimates are:

  • The estimates [6] in are based on (estimated) date of symptom onset and should, in theory, be less susceptible to issues around testing (such as public holidays for example) than basing it on specimen received date. However, it may be difficult to obtain accurate symptom onset date.
  • Similarly estimates are also made based on admissions and deaths in [6]. It’s very useful to look at estimates based on multiple end points, but in this case these require data sources not accessible to the public.
  • The assumed generation interval assumptions are not exactly the same, but are reasonably similar.

7 Author

This report was prepared by Louis Rossouw. Please get in contact with Louis Rossouw if you have comments or wish to receive this regularly.

Louis Rossouw
Head of Research & Analytics
Gen Re | Life/Health Canada, South Africa, Australia, NZ, UK & Ireland
Email: LRossouw@GenRe.com Mobile: +27 71 355 2550

The views in this document represents that of the author and may not represent those of Gen Re. Also note that given the significant uncertainty involved with the parameters, data and methodology care should be taken with these numbers and any use of these numbers.

References

[1] A. Cori, N. M. Ferguson, C. Fraser, and S. Cauchemez, “A new framework and software to estimate time-varying reproduction numbers during epidemics,” American Journal of Epidemiology, vol. 178, no. 9, pp. 1505–1512, Sep. 2013, doi: 10.1093/aje/kwt133. [Online]. Available: https://doi.org/10.1093/aje/kwt133

[2] A. Cori, EpiEstim: A package to estimate time varying reproduction numbers from epidemic curves. 2013 [Online]. Available: https://CRAN.R-project.org/package=EpiEstim

[3] V. Marivate et al., “Coronavirus disease (COVID-19) case data - South Africa.” Zenodo, 21-Mar-2020 [Online]. Available: https://zenodo.org/record/3888499. [Accessed: 26-Oct-2020]

[4] National Institute for Communicable Diseases, “National COVID-19 Daily Report,” 2021 [Online]. Available: https://www.nicd.ac.za/diseases-a-z-index/covid-19/surveillance-reports/national-covid-19-daily-report/

[5] OCHA, “South africa - subnational administrative boundaries,” Dec. 2018 [Online]. Available: https://data.humdata.org/dataset/south-africa-admin-level-1-boundaries

[6] National Institute for Communicable Diseases, “The Daily COVID-19 Effective Reproductive Number (R) in South Africa,” Week 11 2020 [Online]. Available: https://www.nicd.ac.za/wp-content/uploads/2021/03/COVID-19-Effective-Reproductive-Number-in-South-Africa-week-11.pdf